Social Capital’s Impact on College Graduation Rates,Debt,and Student Loan Defaults

Abstract

This study extends the literature on education economics and student retention by examining social capital as a predictor of college graduation rates, student debt levels, and student loan default rates. Coleman’s social capital theory is employed to understand how social influences can impact students through external social support (i.e., social capital). The study uses school-level data from the U.S. Department of Education’s Integrated Postsecondary Education Data System and two social capital measures. Results suggest that social capital, at both the state and the community level, significantly influences graduation rates, student debt levels, and loan default rates. Implications for theory and practice are discussed.

Keywords

student debt student loan default rates graduation rates social capital

Increased access to higher education has been an important social and political goal for several decades (Association of Public & Land-Grant Universities, 2018; Thomas et al., 1979). Simultaneously, researchers have diligently investigated potential personal, environmental, and socioeconomic factors that may help improve student success and outcomes (for a review, see Van der Zanden et al., 2018).

Apart from increasing access, student loans continue to be a major concern in higher education. As of December 2019, the total outstanding student loan debt in the United States was estimated at over $1.6 trillion (Nykiel, 2019). Although several key risk factors at the personal and school levels have been identified by other researchers, few efficient default prevention strategies have been established (Ishitani & McKitrick, 2016; Mueller & Yannelis, 2017).

This study was designed to expand on what is known about current environmental factors related to graduation rates, student debt levels, and student loan default rates. This article uses school-level data from the U.S. Department of Education’s Integrated Postsecondary Education Data System (IPEDS), regional economic data from the U.S. Bureau of Economic Analysis, and two measures of social capital to test the influences of external social support on student outcomes. Previously, researchers have demonstrated strong support for individual social capital in improving student outcomes (e.g., Fernandez-Martinez et al., 2017); however, studies that seek to understand the effects of community social capital are lacking.

Recently, researchers have considered various noncognitive influences on academic success. Farruggia et al. (2018) elucidated how academic mindsets and perseverance predict first-semester grade point average and first-to-second-year retention. Perseverance, or persistence, and the relationship between the student and institution are equally important (Savage et al., 2017). Critically, the public health literature has shown that these traits are positively related to one’s environmental (i.e., community) social capital resources (Aldrich & Meyer, 2015).

More than a decade has passed since Tinto (2006) asked “What next?” concerning the research and practice of student retention. This study seeks to answer this call by studying not only newly identified influences on graduation rates but also the anxiety-inducing concepts of student debt (Archuleta et al., 2013). In addition, regional differences have been found to explain variation in graduation rates and student debt levels without offering explanations for these differences (Mohundro et al., in press). The results of this study suggest that the county and state levels of social capital, as measured using the Putnam index (Putnam, 2000) and the RGF index (Rupasingha et al., 2006), are statistically and economically significant predictors of graduation, access to debt, and default rates. Furthermore, the analyses are robust to multiple alternative tests.

The remainder of the article is organized as follows: the remainder of this section reviews recent social capital and student success literature. The next section describes the sample and methods, with the subsequent section presenting and discussing the results. The final section describes the major conclusions of this study, including known limitations and opportunities for future research.

In an era of exponential revenue growth, higher education has become a veritable marketplace where producers (i.e., colleges and universities) vie for consumers (i.e., students) (Winston, 2000). In an era of rapidly increasing costs, colleges are spending more than ever to attract new students, often taking to television commercials to tout the benefits of the institutions. For example, Southern New Hampshire University, a nonprofit university, spent more than $139 million on television advertising in 2018 alone (McKenzie, 2019).

Apart from advertising, colleges and universities are spending enormous financial resources in other areas to attract and retain students. Hurst (2016) called such an initiative an “arms race” (p. 1). Colleges across the United States have spent more, proportionally, on buildings and amenities since the 2008 to 2009 recession than in any other period in American history (Paterson, 2018). From 2010 to 2018, public and private 4-year colleges in the United States added an average of $128.7 million worth of buildings and land improvements per institution (IPEDS, 2019). During the same period, the average increase in student enrollment was only 264.1 students per institution (IPEDS, 2019), such that one can argue that these building sprees are not in response to increased enrollment.

The question then for researchers and policy makers is what effect these enormous investments are having on student well-being and graduation rates. Colleges are focusing on modern buildings, exciting extracurricular activities, and a steady stream of new construction to lure and retain students even more than quality education metrics (Roberts & Taylor, 2016). Do these expenditures, which are undoubtedly financed by the widely criticized rise in attendance costs, result in successful students?

To answer this question, this study turns to the public health literature to understand how infrastructure interacts with community. Aldrich and Meyer (2015) investigated the effects of rebuilding physical infrastructure after natural disasters and war. The authors found that governments and relief agencies typically focus on rebuilding homes and levees, among others, but social infrastructure drives resilience at a higher rate. Communities with high levels of social infrastructure (i.e., social networks) experience high levels of community resilience, which in turn builds and strengthens individual resilience (Aldrich & Meyer, 2015).

Social networks’ impact on resilience has a direct relationship with multiple positive mental health consequences. Commitment, quality of attention, self-efficacy, and creativity are a few of the psychological well-being perquisites produced by resilience (Fernandez-Martinez et al., 2017). Students with high levels of resilience have high levels of academic achievement due to increased engagement (i.e., defined as work-related, positive, and persistent affective state) and buffered burnout (Fernandez-Martinez et al., 2017).

Schwartz et al. (2018) presented a more explicit link between social capital and college success for first-generation students. They showed that a lack of on-campus connections severely impacts the views and outcomes of these first-generation scholars. Furthermore, such a need for social connections can be developed through intervention when environmental social capital is available (Schwartz et al., 2018).

Researchers have recently highlighted the importance of social capital to diversity in colleges and universities. Whether these studies were in the context of Historically Black Colleges and Universities (e.g., Longmire-Avital & Miller-Dyce, 2015) or in the context of Hispanic students (e.g., Montalvo, 2013; Tovar, 2015; Wagner, 2015), the results consistently show that social capital is critical to student retention and perceived social status. Diversity researchers have also investigated social capital as an explanation for differences in college enrollment. For example, Klevan et al. (2016) examined gender differences in collegiate enrollment, but they did not look at graduation rates. In contrast to the findings of those researchers, who focused their efforts on the role of social networks, involvement within student groups, and interactions with faculty and advisors, the focus of this research is on the impact of existing environmental social capital.

The need for social network support appears to be growing among college students. As population generation’s cycle through undergraduate institutions, the importance of social capital increases. The current generation (i.e., Generation Z, born from 1995 to 2015) has demonstrated the need to be connected to social pipelines and for collaboration with peers (Lindbeck & Fodrey, 2010). Millenials, especially individuals born near the end of the generation era, view college as more of a social capital investment (e.g., meeting new people and the college experience) than a human capital investment (e.g., making more money and finding better job options) (Johnson et al., 2016). This insight is alarming when one considers that millenials also view student loans as investments instead of debts (Chudry et al., 2011).

Simultaneously, millenials are less satisfied with returns on student loans than previous generations who attended college to increase their human capital instead of social capital (Johnson et al., 2016). However, the extent to which community social capital can fill the psychological need for personal social capital development, which may lead to improved student loan outcomes by enabling students to focus their education efforts on postgraduation needs (i.e., increasing skills and knowledge), remains unclear.

This study contributes to the literature on social capital and student success by considering the causal relationship from a different standpoint. Prior researchers have viewed social capital as a resource to be developed within colleges. This article examines whether students can draw upon preexisting environmental social capital in their geographic region. Reinforcing this focus on social capital development within the institution, Lei et al. (2011) highlighted the role of the cohort education model. In the model, being involved in an active community of learners may improve graduation rates. In addition, by developing a social network within the cohort, students can freely exchange information with one another, encourage learning, and mutually promote success. To reiterate, this focus is on the development of social capital, that is, an existing pool of social capital can be beneficial to students throughout their academic careers.

Beyond academic success, the literature has shown that social capital has significant impacts on access to debt financing as well. Mohundro et al. (in press) found that banks are more lenient with lending and interest rate decisions when social capital is high. International researchers have also shown that households in high social capital areas (i.e., areas with low corruption beliefs, high confidence in institutions and authorities, and high prevalence of religiosity) have low loan default rates and fewer late payments (Georgarakos, 2015).

Social capital theory provides the theoretical framework to extend the foregoing literature in this study (Coleman, 1988). Social capital theory posits that social relationships are resources that yield economic benefits. To this end, a model is estimated to test whether environmental social capital, proxied by state and county-level social capital measures, directly impacts student outcomes. The following hypotheses are thus presented.
Hypothesis 1: Social capital predicts 4-year graduation rates, such that the 4-year graduation rate of students from U.S. colleges and universities are high when social capital is high.

Hypothesis 2: Social capital predicts student debt levels, such that the debt levels of students from U.S. colleges and universities are high when social capital is high.

Hypothesis 3: Social capital predicts student loan default rate, such that the loan default rate of students from U.S. colleges and universities is low when social capital is high.

Method

Variables

Social capital is intangible, and the literature has suggested different measures. At its core, however, social capital is derived by networks that enable individual success through cooperation and community trust (Fuyukama, 1997; Guiso et al., 2004). Following Huang and Shang (2019), this study considers society-level mutual trust as the principal social capital proxy using two measures.

The first measure of social capital is the RGF index, a county-level metric developed by Rupasingha et al. (2006) and available from the Northeast Regional Center for Rural Development (2020) at the Pennsylvania State University College of Agricultural Sciences. The authors used principal component analysis to create an index derived from the number of social/civic organizations, presidential election voter turnout, the number of nongovernment organizations, and census response rate for each county. As the metric is reported as an index value, it behaves as a ratio variable where the difference between observed values in different counties is proportional and meaningful. Importantly, the index is derived from an economic maximization model in which each individual is assumed to rationally pursue their best interests. Thus, any decision to participate in civic, cultural, or political activities is done so at the expense of other activities.

This index value is available for the years 1997, 2005, 2009, and 2014. For the intervening years, it is assumed that the index value remains unchanged and the last reported value is used (Huang & Shang, 2019). Figure 1 depicts the RGF social capital index by U.S. county.

Figure 1.
RGF Index Map.

The second social capital measure used is the Putnam index (Putnam, 2000), which is a state-level index. The Putnam index uses 14 state-level social capital metrics obtained through surveys. These metrics include volunteering, entertaining in the home, club/organization meeting attendance, community service, time spent visiting friends, club/organization leadership roles, view of community honesty, number of civic/social organizations, and voter turnout. These data are available on Robert D. Putnam’s website (Putnam, 2019). In contrast to the RGF index, the Putnam index is a state-level (i.e., not a county-level) cross-sectional index and includes regional trust measures. Figure 2 displays the Putnam social capital index by U.S. state.

Figure 2.
Putnam Index Map.

The county- and state-level social capital metrics are especially important because college graduation rates and student loan outcomes are significantly influenced by geographic region (Mohundro et al., in press). Following that study, geographic region fixed effects are controlled using the regions designated by the U.S. Bureau of Economic Analysis (2019a, 2019b).

Control variable data were collected from IPEDS (2019) of the U.S. Department of Education.

Sample

The original sample was composed of 6,175 colleges and universities. A total of 441 schools that did not report the school name and 130 international schools were omitted. Furthermore, the study excluded 2,881 schools that did not report graduation rates because the study investigates the influence of regional social capital on 4-year graduation rates. In addition, 519 for-profit schools were excluded because their student population is disproportionately outside of the school’s home region. Finally, 334 cases were removed that did not report student debt or enrollment information. The final sample consisted of 1,870 valid and consistent cases. However, the sample was further reduced for certain analyses (e.g., the Putnam index did not capture the state of Alaska, the RGF index was unavailable for certain institution locations, and several control variables did not have data available for every institution). Finally, student default rates were unavailable for 505 institutions, which further limited the sample for certain analyses.

Descriptive Statistics

Table 1 presents the summary statistics for (Panel A) and Pearson’s correlations between (Panel B) all variables used in the study. All calculations were conducted in R version 3.6.0. As shown in Panel A, the average salary of a professor was $83,899 for the sample with a standard deviation of $25,423. This study uses this value as a proxy for school quality when controlling for the cost of living differences by controlling for regionality. The rationale is that schools are better because they have high-quality professors who can command high compensation on the open market. The average school enrollment is 5,136 with a relatively large standard deviation of 6,921. In the sample, 63.3% of the schools are private with the remainder being public institutions (for-profit colleges and trade schools were excluded from the sample). Standardized average SAT scores were used as a proxy for student quality. In addition, 4-year graduation rates were standardized. The average student debt level at graduation was $6,800 with a standard deviation of $1,556. Finally, student default rate displayed a mean of 7.7% with a relatively large standard deviation of 5.7%. The univariate correlations (Panel B) support our hypotheses, that is, graduation rates and student debt levels are positively correlated with the Putnam and RGF social capital indices, whereas the indices are negatively correlated with student loan defaults. Other correlations are as expected and supported the findings of the prior literature.

Table 1.
Summary Statistics and Correlation Matrix.

Panel A: summary statistics

Variable Proxy Mean Median Standard deviation N

Social capital indices

State level Putnam index −0.143 −0.190 0.617 1,865

County level RGF index −0.319 −0.351 0.752 1,867

School characteristics

School quality Avg. professor salary ($) 83,899 81,634 25,423 1,781

School size Enrollment 5,136 2,276 6,921 1,870

School type Private = 1, public = 0 0.633 1.000 0.482 1,870

Student characteristics

Student quality Average SAT 0.000 −0.280 1.000 1,825

4-year graduation rate Average 0.000 −0.145 1.000 1,870

Student debt Average at graduation ($) 6,800 6,768 1,556 1,870

Student default Average (%) 7.70 6.40 5.71 1,365

Panel B: correlations

Variable Putnam index RGF index School quality School size Private school Student quality Graduation rate Student debt Student default

Putnam index 1.000

RGF index .362 1.000

School quality .035 −.038 1.000

School size −.046 −.158 .407 1.000

Private school .014 .063 −.165 −.558 1.000

Student quality .123 .036 .664 .206 .110 1.000

4-year grad. rate .097 .119 .509 −.067 .399 .716 1.000

Student debt .126 .131 .078 −.183 .412 .089 .355 1.000

Student default −.183 −.050 −.489 −.040 −.216 −.635 −.629 −.286 1.000

Note. SAT = Scholastic Aptitude Test.

Table 1 reports summary statistics (Panel A) on the main variables used in the regression analyses throughout the paper and correlation matrix (Panel B) between them. Putnam index is the U.S. state-level social capital index, and RGF index is the U.S. county-level social capital index. 4-year grad rate denotes the standardized average 4-year graduation rate of students in U.S. colleges and universities. Student debt is the average student debt at graduation by institution for students graduating from U.S. colleges and universities. School quality is proxied by the average professor salary by institution for U.S. colleges and universities in hundreds of thousands of dollars. Student quality is proxied by the standardized average student SAT score per institution. School size is measured by institution enrollment. Private school is a binary variable coded as 1 for private institutions and 0 for public institutions.

Results

Analysis

Ordinary least squares (OLS) regression was used for testing all hypotheses using R version 3.6.0. Transformation of control variables was employed, as required, to enable linearity. OLS assumptions were met except for limited correlations (variance inflation factor <5) between independent variables that did not substantially impact the models.

For simplicity and brevity, the model equation is specified generally in the following form:
$\begin{array}{l} Dependent variable = β_{0} + β_{1} \times social capital index \\ + β_{2} \times school characteristic matrix \\ + β_{3} \times student characteristic matrix \\ + β_{4} \times regions matrix + ε \end{array}$

The dependent variable differs by hypothesis. For Hypothesis 1, the dependent variable is 4-year graduation rates. For Hypotheses 2 and 3, the dependent variables are student loan amounts and student default rates, respectively.

Empirical Results

Effects of Social Capital on Graduation Rates

As previously discussed, the first hypothesis is that social capital will predict 4-year graduation rates, such that the 4-year graduation rate of students from U.S. colleges and universities are high when social capital is high, with the opposite also being true. To assess this hypothesis, a series of OLS regressions was conducted. The dependent variable is 4-year graduation rates, with the Putnam and RGF measures of social capital as the independent variables. The control variables follow the prior work of Mohundro et al. (2019). Table 2 provides the results of the OLS regression and discusses specific control variables and transformations.

Table 2.
Effect of Social Capital on 4-Year Graduation Rates.

Dependent variable
4-year graduation rate

(1) (2) (3) (4)

Intercept 0.0255 (1.077) −49.4100*** (−4.237) 0.0534* (2.141) −44.2600* (−3.784)

Putnam index 0.1568* (4.195) 0.0798* (2.025)

RGF index 0.1606* (5.255) 0.0583 (3.007)

Student quality 0.5629* (29.463) 0.5634* (29.61)

Student debt 1.0670* (3.982) 9.5060* (3.538)

Student debt² −0.5767* (−3.748) −0.5107* (−3.311)

School quality 0.5794* (6.310) 0.5686* (6.229)

School size −0.0481* (−3.315) −0.0445 (−3.063)

Private × School Size 0.0753* (15.377) 0.0758* (15.550)

Region (New England = 0)

Southwest −0.3158* (−4.067) −0.3343* (−4.659)

Southeast −0.0480 (−0.647) −0.1269* (−2.252)

Mountain −0.3340* (−3.739) −0.3055* (−3.411)

Northeast 0.0738 (1.189) 0.0309 (0.553)

West −0.3105* (−4.835) −0.3272* (−5.231)

Midwest −0.2091* (−3.432) −0.2334* (−4.012)

Plains −0.2532* (−3.865) −0.2399* (−3.747)

Observations 1,864 1,775 1,866 1,777

R ² .0094 .6897 .0146 .6898

Adj. R² .0088 .6872 .0141 .6873

F statistic 17.60 279.60 27.61 280.00

F statistic df 1 and 1,863 14 and 1,761 1 and 1,865 14 and 1,763

Note. t-statistics are reported parenthetically below each coefficient.

p < .05. p < .01. p < .001.

Table 2 reports the results from OLS regressions that relate two measures of social capital to 4-year graduation rates of students at U.S. colleges and universities after controlling for other known predictors of 4-year graduation rates. 4-year graduation rate is standardized. Student debt is the natural log of average student debt by institution at graduation with student debt² being student debt squared. Student quality stands for the standardized average student SAT score per institution. School size is the natural log of institution enrollment school quality denotes the average professor salary by institution in hundreds of thousands of dollars. Other variables are not transformed and are described earlier.

Whether using the state-level social capital index (Putnam) or community-level social capital index (RGF) and whether or not all known control variables are included, social capital positively predicts the 4-year graduation rates of U.S. college and university students. This finding is indicated in Table 2 by showing a positive and statistically significant coefficient for both of the social capital measures for all four analyses. In other words, when social capital is high, students are more likely to graduate in the standard 4-year time frame. Column 2 demonstrates this finding using the Putnam index and all known controls (β = 0.0798; p = .0430). Column 4 demonstrates this finding using the RGF index and all known controls (β = 0.0583; p = .0027). Although the study did not assess causation, this is potentially due to societal pressures inherent in social capital (Poland, 2000). Thus, the results support Hypothesis 1.

Effects of Social Capital on Student Debt Amount

As previously discussed, the second hypothesis is that social capital will predict student debt levels, such that the debt levels of students from U.S. colleges and universities are high when social capital is high, with the opposite also being true. To test this hypothesis, a series of OLS regressions was conducted. The dependent variable is student debt levels at graduation, whereas the Putnam and RGF measures of social capital are the independent variables. Once again, the control variables follow the prior work of Mohundro et al. (in press). Table 3 provides the results and discusses the specific control variables and transformations.

Table 3
Effect of Social Capital on Student Debt Amounts at Graduation.

Dependent variable
Student debt amount at graduation

(1) (2) (3) (4)

Intercept 6,847.6* 3,915.0* 6,886.5* 4,081.0*

(186.450) (12.032) (177.456) (12.835)

Putnam index 317.8* 332.1*

(5.479) (3.577)

RGF index 262.1* 174.8*

(5.514) (3.848)

Student quality −173.0* −162.3*

(−3.826) (−3.604)

School quality 488.6* 405.5

(2.262) (1.891)

Private school 1743.0* 1726.0*

(19.905) (19.782)

Region (New England = 0)

Southwest −862.3* (−4.728) −999.0* (−5.956)

Southeast −706.1* (−4.052) −1,064.0* (−8.179)

Mountain −1,010.0* (−4.817) −927.4* (−4.412)

Northeast −138.4 (−0.944) −329.6* (−2.505)

West −1,221.0* (−8.21) −1,274.0* (−8.844)

Midwest −591.2* (−4.143) −717.3* (−5.298)

Plains −779.7* (−5.092) −707.0* (−4.717)

Observations 1,864 1,775 1,866 1,777

R ² .0159 .2801 .0160 .2802

Adj. R² .0153 .2752 .0155 .2753

F statistic 30.02 57.18 30.40 57.26

F statistic df 1 and 1,863 12 and 1,763 1 and 1,865 12 and 1,765

Note. t-statistics are reported parenthetically below each coefficient.

p < .05. p < .01. p < .001.

Table 3 reports the results from OLS regressions in relation to two measures, namely, social capital and student debt amounts at graduation of students in U.S. colleges and universities after controlling for other known predictors of student debt amounts. Variable definitions remain unchanged from Table 2.

Similar to the results of the analysis to assess Hypothesis 1, whether using the Putnam or RGF index with or without controls, social capital positively predicts student loan amounts for students in U.S. colleges and universities. This finding is indicated in Table 3 by showing a positive and statistically significant coefficient for both of the social capital measures for all four analyses. In other words, when social capital is high, students attain more debt. Column 2 in Table 3 demonstrates this finding using the Putnam index and all known controls (β = 332.1; p = .0004). Column 4 in Table 3 presents this finding using the RGF index and all known controls (β = 174.8; p = .0001). Again, causation is not assessed, though it may be due to students’ increased access to debt (Du et al., 2015) and possibly due to a desire to not be a financial burden on their parents (Ruef & Kwon, 2016). Both factors are consistent with high levels of social capital. Therefore, the results fully support Hypothesis 2.

Effects of Social Capital on Student Loan Default

As previously discussed, the final hypothesis is that social capital will predict student loan default rate, such that the loan default rate of students in U.S. colleges and universities will be low when social capital is high, with the opposite also being true. To assess this hypothesis, a series of OLS regressions was conducted. The dependent variable is student loan default rate, whereas the Putnam and RGF measures of social capital are the independent variables. Once again, the control variables follow the prior work by Mohundro et al. (2019). Table 4 presents the results and discusses the specific control variables and transformations.

Table 4.
Effect of Social Capital on Student Loan Default Rates.

Student loan default rate

Dependent variable (1) (2) (3) (4)

Intercept 7.416* (47.258) 52.250* (11.965) 7.566* (45.663) 53.170* (12.147)

Putnam index −1.731* (−7.068) −0.544 (−2.974)

RGF index −0.4177* (−2.035) −0.192 (−1.293)

Student debt −3.982* (−7.544) −4.072* (−7.69)

Student quality −3.317* (−19.957) −3.375* (−20.537)

School quality −10.456* (−10.456) −19.790* (−10.598)

School quality² 9.368* (10.014) 9.572* (10.238)

Private −1.206* (−4.184) −1.201* (−4.171)

Observations 1,357 1,318 1,361 1,322

R ² .0355 .5108 .0030 .5069

Adj. R² .0348 .5086 .0023 .5046

F statistic 49.96 228.30 4.14 225.40

F statistic df 1 and 1,356 6 and 1,312 1 and 1,360 6 and 1,316

Note. t-statistics are reported parenthetically below each coefficient.

p < .05. p < .01. p < .001.

Table 4 reports the results from OLS regressions in relation to two measures, namely, social capital and student loan default rates of students in U.S. colleges and universities after controlling for other known predictors of student loan default rates. Variables are consistent with Tables 2 and 3.

Contrary to the case of the results for Hypotheses 1 and 2, the current results are mixed. At the state level (using the Putnam index as the measure of social capital), as indicated in Column 2, our hypothesis is supported (β = −0.544; p = .0030). This negative and statistically significant coefficient indicates that social capital negatively predicts student loan default rates, as expected. However, at the county or community level (RGF index), the finding is not statistically significant when the proper controls are employed, as shown in Column 4 (β = −0.192; p = .1960). Interestingly, the region matrix employed in prior analyses was also statistically insignificant, as pointed out in Mohundro et al. (in press), although this analysis is not included for brevity. The impact of social capital on student default rate, therefore, is a noncommunity-based effect and requires further research. Although this model does not assess causation, the effect of social capital on default rates is consistent with the literature on personal bankruptcies (Agarwal et al., 2011) and other loan defaults (Georgarakos, 2015). Hypothesis 3 is only supported at the U.S. state level but not the county level.

Discussion

Social capital, as proxied by two distinct measures used frequently in the literature, is a statistically and economically significant predictor of college graduation rates, access to student debt, and student loan default rates. Through county-level (RGF index; Rupasingha et al., 2006) and state-level (Putnam index; Putnam, 2000) environmental social capital indices, the results suggest that college students can internalize these resources to improve their educational outcomes.

The study provides important contributions to the literature on education economics and student retention. First, it identifies an external resource that reduces student loan default rates. This point is critical as researchers and policy makers continue to seek solutions to the country’s student debt crisis. Previous studies have shown strong links between developing student social capital and success, and this study extends this stream of research by demonstrating that these effects can be achieved through community networks as well.

Researchers have very recently demonstrated that regional differences impact graduation rates, student debt levels, and student loan default rates (Mohundro et al., in press). However, these researchers made no attempt to understand why these differences exist. This research findings provide compelling evidence that social capital is one of such antecedents. Future researchers can provide important insight into these questions through further topic development.

In addition, the conclusions drawn in this study build on the findings of other researchers on regional economic development and demonstrate that certain areas with high social capital, in fact, produce high graduation rates. Conclusions by other researchers have demonstrated that regional studies are important because they can identify the best strategies for economic development (Horlings & Padt, 2013). The societal impacts of regional studies have also shown that one of the key factors in developing learned communities is through knowledge exchange and dissemination (Hopkins, 2011). Previously, regions have primarily been utilized as controls. However, this study expounds on recent research (Mohundro et al., in press), which focus on regional differences in graduation rates. Furthermore, the study provides an explanation why such a case exists. The study shows that areas with high levels of social capital result in high graduation rates. Students enrolled in these regions (e.g., states in the Midwest) are more actively involved in their immediate social networks and within their community (Rupasingha et al., 2006). This notion might be due to the fact that high levels of college graduation rates can increase local education (Winters, 2018). Areas with higher education have already been shown to produce areas with higher social capital (Rupasingha et al., 2006). Further research is required in this area; however, the possible explanation is a positive educational feedback loop. As students graduate from college, they obtain jobs within the local area, become active members of the community, and become a resource for others to draw on as they attend college and subsequently graduate.

This study is not without limitations. The cross-sectional nature of the Putnam index does not allow us to rule out endogeneity or understand causal directions. This study attempts to reduce the likelihood of endogeneity bias using regional controls, which have recently been shown to be related to the three dependent variables of the study. The second limitation of the variables is the infrequent updates of the RGF index. Future research should include time-series data or student sentiment surveys to better develop causal analysis.

Arguably, the proxy variables may inadequately capture the intended measures. Although professor salaries as a proxy for school quality and average incoming SAT scores as a proxy for student quality have been used in the previous work, it must be acknowledged that these factors may be imperfect proxies for an entire undergraduate institution. Future work can use finer grained variables in a small sample to test the appropriateness of such measures. In summary, although these proxies are imperfect, they have been used extensively throughout the literature and allow us to control for potential alternative explanations when testing for the main effects.

Perhaps the most important limitation is the lack of a true social capital measure. The Putnam and RGF indices are constructs developed to predict (but not actually measure) the social capital of a person living within a regional boundary (i.e., a U.S. state [Putnam] or county of a U.S. state [RGF]). Not everyone within a regional boundary, whether a state or county, exhibits identical behaviors as others in the same region. As such, the results of this study and many other academic studies demonstrate the reliability of these constructs. However, this study assessed the regional/state and community social capital to determine whether the effects noted were regional or community effects. In addition, the findings are dependent on the research question posed, that is, whether or not graduation rates and access to and usage of student debt are predicted by community and regional social capital, whereas student loan default rates are not driven by community social capital, but only social capital at the broader state level.

Finally, as in all regression analyses, the omitted variables can introduce bias, but all known predictor variables identified in the prior literature were included in this study to mitigate the bias source. However, further research can assess not only the county or state from which a student graduates from an institution of higher learning but also where that student was born or from where that student completed their primary education.

This study merges the understanding of the literature on public health of environmental social capital with the knowledge of the literature on student retention about resilience-driven success. Thus, this study encourages future research in this stream to clarify the causal direction and better understand the link between community social capital- and intervention-based social capital (e.g., student group membership) and the causal relationship between reduced student loan defaults and community trust.

Panel A: summary statistics
Social capital indices
State level	Putnam index	−0.143	−0.190	0.617	1,865
County level	RGF index	−0.319	−0.351	0.752	1,867
School characteristics
School quality	Avg. professor salary ($)	83,899	81,634	25,423	1,781
School size	Enrollment	5,136	2,276	6,921	1,870
School type	Private = 1, public = 0	0.633	1.000	0.482	1,870
Student characteristics
Student quality	Average SAT	0.000	−0.280	1.000	1,825
4-year graduation rate	Average	0.000	−0.145	1.000	1,870
Student debt	Average at graduation ($)	6,800	6,768	1,556	1,870
Student default	Average (%)	7.70	6.40	5.71	1,365

Panel B: correlations
Putnam index	1.000
RGF index	.362	1.000
School quality	.035	−.038	1.000
School size	−.046	−.158	.407	1.000
Private school	.014	.063	−.165	−.558	1.000
Student quality	.123	.036	.664	.206	.110	1.000
4-year grad. rate	.097	.119	.509	−.067	.399	.716	1.000
Student debt	.126	.131	.078	−.183	.412	.089	.355	1.000
Student default	−.183	−.050	−.489	−.040	−.216	−.635	−.629	−.286	1.000

Dependent variable	4-year graduation rate
Intercept	0.0255 (1.077)	−49.4100*** (−4.237)	0.0534* (2.141)	−44.2600*** (−3.784)
Putnam index	0.1568*** (4.195)	0.0798* (2.025)
RGF index			0.1606*** (5.255)	0.0583** (3.007)
Student quality		0.5629*** (29.463)		0.5634*** (29.61)
Student debt		1.0670*** (3.982)		9.5060*** (3.538)
Student debt²		−0.5767*** (−3.748)		−0.5107*** (−3.311)
School quality		0.5794*** (6.310)		0.5686*** (6.229)
School size		−0.0481*** (−3.315)		−0.0445** (−3.063)
Private × School Size		0.0753*** (15.377)		0.0758*** (15.550)
Region (New England = 0)
Southwest		−0.3158*** (−4.067)		−0.3343*** (−4.659)
Southeast		−0.0480 (−0.647)		−0.1269* (−2.252)
Mountain		−0.3340*** (−3.739)		−0.3055*** (−3.411)
Northeast		0.0738 (1.189)		0.0309 (0.553)
West		−0.3105*** (−4.835)		−0.3272*** (−5.231)
Midwest		−0.2091*** (−3.432)		−0.2334*** (−4.012)
Plains		−0.2532*** (−3.865)		−0.2399*** (−3.747)
Observations	1,864	1,775	1,866	1,777
R ²	.0094	.6897	.0146	.6898
Adj. R²	.0088	.6872	.0141	.6873
F statistic	17.60	279.60	27.61	280.00
F statistic df	1 and 1,863	14 and 1,761	1 and 1,865	14 and 1,763

Dependent variable	Student debt amount at graduation
Intercept	6,847.6***	3,915.0***	6,886.5***	4,081.0***
(186.450)	(12.032)	(177.456)	(12.835)
Putnam index	317.8***	332.1***
(5.479)	(3.577)
RGF index			262.1***	174.8***
		(5.514)	(3.848)
Student quality		−173.0***		−162.3***
	(−3.826)		(−3.604)
School quality		488.6*		405.5
	(2.262)		(1.891)
Private school		1743.0***		1726.0***
	(19.905)		(19.782)
Region (New England = 0)
Southwest		−862.3*** (−4.728)		−999.0*** (−5.956)
Southeast		−706.1*** (−4.052)		−1,064.0*** (−8.179)
Mountain		−1,010.0*** (−4.817)		−927.4*** (−4.412)
Northeast		−138.4 (−0.944)		−329.6* (−2.505)
West		−1,221.0*** (−8.21)		−1,274.0*** (−8.844)
Midwest		−591.2*** (−4.143)		−717.3*** (−5.298)
Plains		−779.7*** (−5.092)		−707.0*** (−4.717)
Observations	1,864	1,775	1,866	1,777
R ²	.0159	.2801	.0160	.2802
Adj. R²	.0153	.2752	.0155	.2753
F statistic	30.02	57.18	30.40	57.26
F statistic df	1 and 1,863	12 and 1,763	1 and 1,865	12 and 1,765

	Student loan default rate
Intercept	7.416*** (47.258)	52.250*** (11.965)	7.566*** (45.663)	53.170*** (12.147)
Putnam index	−1.731*** (−7.068)	−0.544** (−2.974)
RGF index			−0.4177* (−2.035)	−0.192 (−1.293)
Student debt		−3.982*** (−7.544)		−4.072*** (−7.69)
Student quality		−3.317*** (−19.957)		−3.375*** (−20.537)
School quality		−10.456*** (−10.456)		−19.790*** (−10.598)
School quality²		9.368*** (10.014)		9.572*** (10.238)
Private		−1.206*** (−4.184)		−1.201*** (−4.171)
Observations	1,357	1,318	1,361	1,322
R ²	.0355	.5108	.0030	.5069
Adj. R²	.0348	.5086	.0023	.5046
F statistic	49.96	228.30	4.14	225.40
F statistic df	1 and 1,356	6 and 1,312	1 and 1,360	6 and 1,316

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Steve Joanis

J. D. Mohundro

Author Biographies

Steve Joanis is Doctor of Business Administration (DBA) student in the Heider College of Business at Creighton University, Omaha, Nebraska, USA.

James Burnley is Doctor of Business Administration (DBA) student in the Heider College of Business at Creighton University, Omaha, Nebraska, USA.

J. D. Mohundro is Doctor of Business Administration (DBA) student in the Heider College of Business at Creighton University, Omaha, Nebraska, USA.

References

Agarwal

Chomsisengphet

Liu

(2011). Consumer bankruptcy and default: The role of individual social capital. Journal of Economic Psychology, 32(4), 632–650. https://doi.org/10.1016/j.joep.2010.11.007

Aldrich

D. P.

Meyer

M. A.

(2015). Social capital and community resilience. American Behavioral Scientist, 59(2), 254–269. https://doi.org/10.1177/0002764214550299

Archuleta

K. L.

Dale

Spann

S. M.

(2013). College students and financial distress: Exploring debt, financial satisfaction, and financial anxiety. Journal of Financial Counseling and Planning, 24(2), 50–62.

Association of Public & Land-Grant Universities. (2018). Why higher education? https://www.aplu.org/projects-and-initiatives/international-programs/past-projects/knowledge-center-for-advancing-development-through-higher-education/why-higher-education.html

Chudry

Foxall

Pallister

(2011). Exploring attitudes and predicting intentions: Profiling student debtors using an extended theory of planned behavior. Journal of Applied Social Psychology, 41(1), 119–149. https://doi.org/10.1111/j.1559-1816.2010.00705.x

Coleman

J. S.

(1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95–S120. https://doi.org/10.1086/228943

Guariglia

Newman

(2015). Do social capital building strategies influence the financing behavior of Chinese private small and medium-sized enterprises? Entrepreneurship Theory and Practice, 39(3), 601–631. https://doi.org/10.1111/etap.12051

Farruggia

S. P.

Han

C-W.

Watson

Moss

T. P.

Bottoms

B. L.

(2018). Noncognitive factors and college student success. Journal of College Student Retention: Research, Theory & Practice, 20(3), 308–327. https://doi.org/10.1177/1521025116666539

Fernandez-Martinez

Andina-Diaz

Fernandez-Pena

Garcia-Lopez

Fulgueiras-Carrill

Liebana-Presa

(2017). Social networks, engagement, and resilience in university students. International Journal of Environmental Research and Public Health, 14(12), 1488–1504. https://doi.org/10.3390/ijerph14121488

10.

Fuyukama

(1997). Social capital and the modern capitalist economy: Creating a high trust workplace. Stern Business Magazine, 4(1), 1–16.

11.

Georgarakos

(2015). Household repayment behavior: The role of social capital and institutional, political, and religious beliefs. European Journal of Political Economy, 37, 249–265. https://doi.org/10.1016/j.ejpoleco.2014.10.001

12.

Guiso

Sapienza

Zingales

(2004). The role of social capital in financial development. American Economic Review, 94(3), 526–556. https://doi.org/10.1257/0002828041464498

13.

Hopkins

(2011). The role of learned societies in knowledge exchange and dissemination: The case of the Regional Studies Association, 1965–2005. History of Education, 40(2), 255–271. https://doi.org/10.1080/0046760X.2010.518161

14.

Horlings

Padt

(2013). Leadership for sustainable regional development in rural areas: Bridging personal and institutional aspects. Sustainable Development, 21(6), 413–424. https://doi.org/10.1002/sd.526

15.

Huang

Shang

(2019). Leverage, debt maturity, and social capital. Journal of Corporate Finance, 54(1), 26–56. https://doi.org/10.1016/j.jcorpfin.2018.11.001

16.

Hurst

(2016). End the extracurricular arms race. Inside Higher Ed. https://www.insidehighered.com/views/2016/04/11/extracurriculars-are-robbing-students-their-education-essay

17.

Integrated Postsecondary Education Data System. (2019). National Center for Education Statistics. U.S. Department of Education.

18.

Ishitani

T. T.

McKitrick

S. A.

(2016). Are student loan default rates linked to institutional capacity? Journal of Student Financial Aid, 46(1), 3.

19.

Johnson

C. L.

Gutter

Cho

S. H.

DeVaney

(2016). Perceived value of college as an investment in human and social capital: Views of generation X and Y. Family & Consumer Sciences Research Journal, 45(2), 193–207. https://doi.org/10.1111/fcsr.12195

20.

Klevan, S., Weinberg, S. L., & Middleton, J. A. (2016). Why the boys are missing: Using social capital to explain gender differences in college enrollment for public high school students. Research in Higher Education, 57(2), 223–257.

21.

Lei, S., Gorelick, D., Short, K., Smallwood, L., & Wright-Porter, K. (2011). Academic cohorts: Benefits and drawbacks of being a member of a community of learners. Education, 131(3).

22.

Lindbeck

Fodrey

(2010). Using technology in undergraduate admission: A student perspective. Journal of College Admission, 208, 10–17.

23.

Longmire-Avital, B. & Miller-Dyce, C. (2015). Factors related to perceived status in the campus community for first generation students at an HBCU. College Student Journal, 49(3), 375–386.

24.

McKenzie

(2019). Marketing for a massive online university. Inside Higher Ed. https://www.insidehighered.com/news/2019/10/08/how-marketing-helped-southern-new-hampshire-university-make-it-big-online

25.

Mohundro

Austin

Weinberger

(2019). Corporate social responsibility and small- and medium-sized firms debt covenants [Manuscript submitted for publication], Creighton University.

26.

Mohundro

Joanis

S. T.

Burnley

(in press). Geographic region, student loans, and college graduation rates. International Journal of Education Economics and Development.

27.

Montalvo, E. J. (2013). The recruitment and retention of Hispanic undergraduate students in public universities in the United States, 2000-2006. Journal of Hispanic Higher Education, 12(3), 237–255.

28.

Mueller

H. M.

Yannelis

(2017). Students in distress: Labor market shocks, student loan default, and federal insurance programs (No. w23284). National Bureau of Economic Research. https://doi.org/10.3386/w23284

29.

Northeast Regional Center for Rural Development. (2020). County-level measure of social capital. Pennsylvania State University. https://aese.psu.edu/nercrd/community/social-capital-resources

30.

Nykiel

(2019). 2019 student loan debt statistics. Nerdwallet. https://www.nerdwallet.com/blog/loans/student-loans/student-loan-debt/

31.

Paterson

(2018). After building spree, college maintenance spending hits 11-year high. Education Dive. https://www.educationdive.com/news/after-building-spree-college-maintenance-spending-hits-11-year-high/544851/

32.

Poland

B. D.

(2000). The ‘considerate’ smoker in public space: The micro-politics and political economy of ‘doing the right thing. Health & Place, 6(1), 1–14. https://doi.org/10.1016/S1353-8292(99)00025-8

33.

Putnam

R. D.

(2000). Bowling alone: The collapse and revival of American community. Simon and Schuster.

34.

Putnam

R. D.

(2019). Research. Bowling Alone: The Collapse and Revival of American Community. http://www.bowlingalone.com

35.

Roberts

Taylor

(2016). The campus matters: Acquiring the competitive edge. Planning for Higher Education, 44(3), 12–18.

36.

Ruef

Kwon

S. W.

(2016). Neighborhood associations and social capital. Social Forces, 95(1), 159–190. https://doi.org/10.1093/sf/sow053

37.

Rupasingha

Goetz

S. A.

Freshwater

(2006). The production of social capital in US counties. The Journal of Socio-Economics, 35(1), 83–101. https://doi.org/10.1016/j.socec.2005.11.001

38.

Savage

M. W.

Strom

R. S.

Ebesu Hubbard

A. S.

Aune

K. S.

(2017). Commitment in college student persistence. Journal of College Student Retention: Research, Theory & Practice, 21(2), 242–264. https://doi.org/10.1177/1521025117699621

39.

Schwartz

S. E.

Kanchewa

S. S.

Rhodes

J. E.

Gowdy

Stark

A. M.

Horn

J. P.

Parnes

Spencer

(2018). I’m having a little struggle with this, can you help me out?” Examining impacts and processes of a social capital intervention for first-generation college students. American Journal of Community Psychology, 61(1–2), 166–178. https://doi.org/10.1002/ajcp.12206

40.

Thomas

G. E.

Alexander

K. L.

Eckland

B. K.

(1979). Access to higher education: The importance of race, sex, social class, and academic credentials. The School Review, 87(2), 133–156. https://doi.org/10.1086/443466

41.

Tinto

(2006). Research and practice of student retention: What next? Journal of College Student Retention: Research, Theory & Practice, 8(1), 1–19.

42.

Tovar, E. (2015). The role of faculty, counselors, and support programs on Latino/a community college students’ success and intent to persist. Community College Review, 43(1), 46–71.

43.

U.S. Bureau of Economic Analysis. (2019a). Gross domestic product by state.

44.

U.S. Bureau of Economic Analysis. (2019b). Regional economic accounts. https://www.bea.gov/data/economic-accounts/regional

45.

Van der Zanden

P. J.

Denessen

Cillessen

A. H.

Meijer

P. C.

(2018). Domains and predictors of first-year student success: A systematic review. Educational Research Review, 23, 57–77. https://doi.org/10.1016/j.edurev.2018.01.001

46.

Wagner, J. M. (2015). Hispanic minority college students at selective colleges: What matters with degree completion? Journal of Hispanic Higher Education, 14(4), 303–326.

47.

Winston

G. C.

(2000). The positional arms race in higher education (No. 54). WPEHE Discussion Paper. Williams College.

48.

Winters

J. V.

(2018). Do higher college graduation rates increase local education levels? Papers in Regional Science, 97(3), 617–638. https://doi.org/10.1111/pirs.12258

Panel A: summary statistics
Variable	Proxy	Mean	Median	Standard deviation	N
Social capital indices
State level	Putnam index	−0.143	−0.190	0.617	1,865
County level	RGF index	−0.319	−0.351	0.752	1,867
School characteristics
School quality	Avg. professor salary ($)	83,899	81,634	25,423	1,781
School size	Enrollment	5,136	2,276	6,921	1,870
School type	Private = 1, public = 0	0.633	1.000	0.482	1,870
Student characteristics
Student quality	Average SAT	0.000	−0.280	1.000	1,825
4-year graduation rate	Average	0.000	−0.145	1.000	1,870
Student debt	Average at graduation ($)	6,800	6,768	1,556	1,870
Student default	Average (%)	7.70	6.40	5.71	1,365